农业大数据学报 ›› 2021, Vol. 3 ›› Issue (3): 13-22.doi: 10.19788/j.issn.2096-6369.210302

• 专题——农业模型 • 上一篇    下一篇

小麦模型WheatSM参数调试优化方法研究

靳霞菲1(), 陈先冠1, 宫志宏2, 冯利平1()   

  1. 1.中国农业大学资源与环境学院,北京 100193
    2.天津市气候中心,天津 300074
  • 收稿日期:2021-07-20 出版日期:2021-09-26 发布日期:2021-12-22
  • 通讯作者: 冯利平 E-mail:1830546228@qq.com;fenglp@cau.edu.cn
  • 作者简介:靳霞菲,女,硕士,研究方向:作物模型调参与优化。E-mail:1830546228@qq.com
  • 基金资助:
    国家重点研发项目(2016YFD0300201);中国气象局河南省农业气象保障与应用技术重点实验室项目(AMF201805)

Parameter Adjustment and Optimization Methods for The WheatSM

Xiafei Jin1(), Xianguan Chen1, Zhihong Gong2, Liping Feng1()   

  1. 1.College of Resources and Environmental Sciences, China Agricultural University, Beijing 100193, China
    2.Tianjin Climate Center, Tianjin 300074, China
  • Received:2021-07-20 Online:2021-09-26 Published:2021-12-22
  • Contact: Liping Feng E-mail:1830546228@qq.com;fenglp@cau.edu.cn

摘要:

作物生长模型是评估作物生产、资源利用及气候变化影响等的有效工具,准确地确定作物模型参数是应用模型的关键。WheatSM(Wheat Growth and Development Simulation Model)模型已在作物生产优化管理上得到一定的应用,并取得较好的效果,但由于该模型参数较多,模型参数调试复杂。为了快速、准确地确定WheatSM模型参数,简化该模型的调参工作,促进其在农业气象领域中广泛应用,本研究在国内外作物模型参数自动调节方法的基础上,基于PEST(Parameter Estimation)方法构建了WheatSM模型参数的自动调节耦合系统,并对WheatSM模型的发育期和产量参数进行了自动寻优。选择北京上庄作为代表性试验点,以试验点的气象数据、土壤数据和2014~2016年冬小麦不同播期试验数据为基础,应用PEST参数自动优化方法和试错法分别对小麦生长模型WheatSM发育期参数和产量参数进行调试,并将优化结果和试错法的模拟结果进行比较。研究结果表明,基于PEST方法的模型参数调节精准度较高,模拟发育期的误差不大于7天,模拟产量的误差不大于228.63kg·hm-2。同时,与试错法相比,PEST方法具有耗时少、可同时批量处理数据、更高效快捷等优点,使用该自动调参系统可减少参数率定的工作量,节省模型的操作时间,简化工作的复杂度和获得较高的模拟精度。该研究为WheatSM模型参数的自动优化提供一种便捷方法,为提高作物模型参数调试的效率和准确性提供了理论参考和指导。

关键词: 小麦, PEST, WheatSM模型, 参数优化, 发育期, 产量, 作物生长模型

Abstract:

The crop growth model is an effective tool to evaluate crop production, resource utilization, and climate change impact. The WheatSM (Wheat Growth and Development Simulation Model) has been applied to crop production optimization and management and has achieved good achievements. However, because of the large number of model parameters, it’s complicated to debug the model parameters. To determine the parameters of the WheatSM model quickly and accurately, it is necessary to simplify the parameter adjustment work of the model and promote its wide application in the field of agricultural meteorology. In this study, on the basis of automatic adjustment methods of crop model parameters at home and abroad, an automatic adjustment coupling system of WheatSM model parameters is constructed based on the PEST (Parameter Estimation) method. The phenology and yield parameters of the WheatSM model were optimized automatically. Shangzhuang, Beijing was selected as a representative site.This study compared the optimization results with the trial-and-error simulation results, andused the automatic optimization method and trial-and-error method to adjust wheat phenology parameters and yield parameters for wheat growth model WheatSM, based on the meteorological data, soil data of the test sites, and the test data of different sowing dates of winter wheat from 2014 to 2016. The results show that the PEST method has high precision and good simulation effect for automatic adjustment and optimization of model parameters. The error of simulated phenology was less than 7 days, and the error of simulated yield was less than 228.63 kg·hm-2. The PEST method has the advantages of being less time consuming and allowing for the simultaneous batch processing of data. Using this automatic parameter adjustment system can reduce the workload of parameter calibration, save model operation time, simplify work complexity, and obtain higher simulation accuracy. This study provides a convenient method for WheatSM parameters automatic optimization and a theoretical reference and guidance for improving the efficiency and accuracy of crop model parameters calibration.

Key words: wheat, PEST, WheatSM model, parameter optimization, phenology, yield, crop growth model

中图分类号: 

  • S152